On-Line Analytical Processing Features

On-Line Analytical Processing - OLAP - is the ability to conduct data analysis without leaving the context of the database. Highlander supports a number of OLAP functions.

Virtual Dataset Measures

Dataset measures can be record counts, summarized field values, or summarized computed values - virtual measures. Virtual Dataset measures are measures created by user-defined calculations. The accompanying illustration is the specification of a virtual measure. Calculations can combine any field in the dataset source table with standard arithmetic operators, built in functions, constants, and custom operators. You can even use custom written DLL's for specialized or proprietary calculations. These calculations are made while the database is loading to create dimensions or dataset values for summarization. Dataset measures must be natively summarized over the dimensions, such as record count, or be supported internally by Highlander. Ratio measures for example are supported if their numerator and denominator are supported. This opens most ratio analysis to Highlander.

Sorting and Censoring of Results

When results are displayed, it is often useful to sort by the value along a dimension, so the important categories in a dimension are easy to find. Select a dimension and make this the x-axis, then click the "Sort Up" or "Sort Down" button on the results toolbar. Compare these displays before and after sorting is applied.

Normalization and Accumulation of Results

It's easy to extract a histogram for the marginal distribution of a variable, given values for other variables. When fixing the other variables at multiple values, you see multiple histograms, but they are not directly comparable since they are essentially based on different sample sizes. Comparisons are possible however, by normalizing each curve so the area under all curves is the same. Check out the two illustrations above. These are the same query results before and after normalization to the constant 100%. Normalization is on the result toolbar. Using the Accumulation feature, data can be integrated over the x-axis dimension. The Accumulation button on the results tool bar integrates the data being displayed. This turns densities for example into cumulative distribution functions, as illustrated below.

Multi-Dimensional Spreadsheets

Multi-dimension query results are viewable in a spreadsheet while retaining visual identification of all dimensions. The accompanying illustration displays a five dimensional result using the spreadsheet rows for two dimensions (Gender, Year), columns for two dimensions (Pay Grade, Designator), and sheets for the remaining dimension (Commissioning Source). Using the spreadsheet axis buttons on the results toolbar, the user can place any data dimension on any spreadsheet axis or rearrange the dimension ordering. Exporting the spreadsheet to Microsoft Excel preserves this multi-dimensional layout.